The Essential Role of Human Databases for Learning in and Validation of Affectively Competent Agents.

This output is an invited and refereed chapter in the second of the two book length outputs resulting from the EU HUMAINE grant and follow-on grants. The book is in the OUP Affective Science Series and is intended to provide a theoretically oriented state of the art model for those working in the area of affective computing. Each chapter provides a synthesis of a specific area and presents new data/findings/approaches developed by the author(s) which take the area further. This chapter is in the section on ‘Approaches to developing expression corpora and databases.’ The chapter provides a critical synthesis of the issues involved in databases for affective computing and introduces the SEMAINE SAL Database, developed as an integral part of the EU SEMAINE Project (The Sensitive Agent Project 2008-2011) which is an interdisciplinary project. The project aimed to develop a computer interface that would allow a human to interact with an artificial agent in an emotional manner.

On-line Emotion Recognition in a 3-D Activation-Valence-Time Continuum using Acoustic and Linguistic Cues

For many applications of emotion recognition, such as virtual agents, the system must select responses while the user is speaking. This requires reliable on-line recognition of the user’s affect. However most emotion recognition systems are based on turnwise processing. We present a novel approach to on-line emotion recognition from speech using Long Short-Term Memory Recurrent Neural Networks. Emotion is recognised frame-wise in a two-dimensional valence-activation continuum. In contrast to current state-of-the-art approaches, recognition is performed on low-level signal frames, similar to those used for speech recognition. No statistical functionals are applied to low-level feature contours. Framing at a higher level is therefore unnecessary and regression outputs can be produced in real-time for every low-level input frame. We also investigate the benefits of including linguistic features on the signal frame level obtained by a keyword spotter.

The development of national in-group bias: English children's attributions of characteristics to English, American and German people

This study investigated the development of national in-group bias in 5-11-year-old children. Three hundred and seven English children were asked to attribute characteristics to their own national group either on its own or in conjunction with attributing characteristics to one of two national out-groups, either Americans or Germans. The importance which the children ascribed to their own national identity in relationship to their other social identities was also assessed. It was found that, with increasing age, there was an increase in the number of negative characteristics attributed to the national in-group, and an increase in the number of positive characteristics attributed to the two out-groups, the net result being an overall reduction in in-group bias across this age range. However, in-group favouritism was still exhibited at all ages. Greater importance was attributed to national identity with increasing age. However, the characteristics attributed to the English in-group did not vary as a function of the comparative out-group which was present while the attributions were being made. The presence of a comparative out-group also did not affect the importance that was ascribed to the national identity. These findings suggest that children are relatively insensitive to the prevailing comparative context when making judgments about national groups.

Young children's evaluations of the ingroup and of outgroups: A multi-national study

In investigating intergroup attitudes, previous research in developmental psychology has frequently confounded ingroup favouritism and outgroup derogation. The present study, using unconfounded measures, examines the possibility that ingroup favouritism and outgroup derogation are distinct phenomena. Six-year-old children (n=594) from five, culturally diverse nations were asked to make various evaluations of the national ingroup and of four national outgroups. The data indicate that although there is overwhelming evidence that young children favour the ingroup over other groups, outgroup derogation is limited in extent and appears to reproduce attitudes held by adult members of the particular nations investigated.

The MATISSE study: a randomised trial of group art therapy for people with schizophrenia

Background: Art Therapy has been promoted as a means of helping people who may find it difficult to express themselves verbally engage in psychological treatment. Group Art Therapy has been widely used as an adjunctive treatment for people with schizophrenia but there have been few attempts to examine its effects and cost effectiveness has not been examined. The MATISSE study aims to evaluate the clinical and cost effectiveness of group Art Therapy for people with schizophrenia.

Childhood Trauma and Hippocampal and Amygdalar Volumes in First-Episode Psychosis

Objective: A history of childhood trauma is common in individuals who later develop psychosis. Similar neuroanatomical abnormalities are observed in people who have been exposed to childhood trauma and people with psychosis. However, the relationship between childhood trauma and such abnormalities in psychosis has not been investigated. This study aimed to explore the association between the experience of childhood trauma and hippocampal and amygdalar volumes in a first-episode psychosis (FEP) population. Methods: The study employed an observational retrospective design. Twenty-one individuals, who had previously undergone magnetic resonance imaging procedures as part of the longitudinal Northern Ireland First-Episode Psychosis Study, completed measures assessing traumatic experiences and were included in the analysis. Data were subject to correlation analyses (rand rob). Potential confounding variables (age at FEP and delay to scan from recruitment) were selected a priori for inclusion in multiple regression analyses. Results: There was a high prevalence of lifetime (95%) and childhood (76%) trauma in the sample. The experience of childhood trauma was a significant predictor of left hippocampal volume, although age at FEP also significantly contributed to this model. There was no significant association between predictor variables and right hippocampal volume. The experience of childhood trauma was a significant predictor of right and total amygdalar volumes and the hippocampal/amygdalar complex volume as a whole. Conclusions: The findings indicate that childhood trauma is associated with neuroanatomical measures in FEP. Future research controlling for childhood traumatic experiences may contribute to explaining brain morphology in people with psychosis.

Risk of Malignant Progression in Barrett’s Esophagus Patients: Results from a Large Population-Based Study

Background: Barrett's esophagus (BE) is a premalignant lesion that predisposes to esophageal adenocarcinoma. However, the reported incidence of esophageal adenocarcinoma in patients with BE varies widely. We examined the risk of malignant progression in patients with BE using data from the Northern Ireland Barrett's esophagus Register (NIBR), one of the largest population-based registries of BE worldwide, which includes every adult diagnosed with BE in Northern Ireland between 1993 and 2005.

Subjects and Methods: We followed 8522 patients with BE, defined as columnar lined epithelium of the esophagus with or without specialized intestinal metaplasia (SIM), until the end of 2008. Patients with incident adenocarcinomas of the esophagus or gastric cardia or with high-grade dysplasia of the esophagus were identified by matching the NIBR with the Northern Ireland Cancer Registry, and deaths were identified by matching with records from the Registrar General's Office. Incidence of cancer outcomes or high-grade dysplasia was calculated as events per 100 person-years (% per year) of follow-up, and Cox proportional hazard models were used to determine incidence by age, sex, length of BE segment, presence of SIM, macroscopic BE, or low-grade dysplasia. All P values were from two-sided tests.

Results: After a mean of 7.0 years of follow-up, 79 patients were diagnosed with esophageal cancer, 16 with cancer of the gastric cardia, and 36 with high-grade dysplasia. In the entire cohort, incidence of esophageal or gastric cardia cancer or high-grade dysplasia combined was 0.22% per year (95% confidence interval [CI] = 0.19% to 0.26%). SIM was found in 46.0% of patients. In patients with SIM, the combined incidence was 0.38% per year (95% CI = 0.31 to 0.46%). The risk of cancer was statistically significantly elevated in patients with vs without SIM at index biopsy (0.38% per year vs 0.07% per year; hazard ratio [HR] = 3.54, 95% CI = 2.09 to 6.00, P <. 001), in men compared with women (0.28% per year vs 0.13% per year; HR = 2.11, 95% CI = 1.41 to 3.16, P <. 001), and in patients with low-grade dysplasia compared with no dysplasia (1.40% per year vs 0.17% per year; HR = 5.67, 95% CI = 3.77 to 8.53, P <. 001).

Conclusion: We found the risk of malignant progression among patients with BE to be lower than previously reported, suggesting that currently recommended surveillance strategies may not be cost-effective. © The Author 2011. Published by Oxford University Press. All rights reserved.

Prognostic Significance of TRAIL Signaling Molecules in Stage II and III Colorectal Cancer

Purpose: We previously found that cellular FLICE-inhibitory protein (c-FLIP), caspase 8, and tumor necrosis factor–related apoptosis-inducing ligand (TRAIL) receptor 2 (DR5) are major regulators of cell viability and chemotherapy-induced apoptosis in colorectal cancer. In this study, we determined the prognostic significance of c-FLIP, caspase 8, TRAIL and DR5 expression in tissues from patients with stage II and III colorectal cancer.

Experimental Design: Tissue microarrays were constructed from matched normal and tumor tissue derived from patients (n = 253) enrolled in a phase III trial of adjuvant 5-fluorouracil–based chemotherapy versus postoperative observation alone. TRAIL, DR5, caspase 8, and c-FLIP expression levels were determined by immunohistochemistry.

Results: Colorectal tumors displayed significantly higher expression levels of c-FLIP (P < 0.001), caspase 8 (P = 0.01), and DR5 (P < 0.001), but lower levels of TRAIL (P < 0.001) compared with matched normal tissue. In univariate analysis, higher TRAIL expression in the tumor was associated with worse overall survival (P = 0.026), with a trend to decreased relapse-free survival (RFS; P = 0.06), and higher tumor c-FLIP expression was associated with a significantly decreased RFS (P = 0.015). Using multivariate predictive modeling for RFS in all patients and including all biomarkers, age, treatment, and stage, we found that the model was significant when the mean tumor c-FLIP expression score and disease stage were included (P < 0.001). As regards overall survival, the overall model was predictive when both TRAIL expression and disease stage were included (P < 0.001).

Conclusions: High c-FLIP and TRAIL expression may be independent adverse prognostic markers in stage II and III colorectal cancer and might identify patients most at risk of relapse.